The most recent strain of infatuation with this rookie journalistic error, from Grits' observation, began a few years ago when federal courts ordered California's prison system to decarcerate because of overcrowding. The papers soon filled with anecdotes of violent crimes that cops and prosecutors attributed to the policy and suggestions that the federal order was responsible for a crime spike. But when the annualized data came out, that trend never materialized. The following year, CA voters eliminated the state's three-strikes law and a similar hue and cry about increased crime ensued. But the annualized data didn't bear that out, either. They're doing the same thing now over Prop 47, which prioritized treatment and changed certain drug crimes to misdemeanors.
The trick seems to be to engage in demagoguery for as long as possible during the period before official numbers are reported, which is a rather long delay. (For example, 2015 UCR data hasn't been reported for Texas yet, so we can for now only talk about data through 2014.) Then, when the numbers come out, if murders go up, critics get to say "I told you so." If they go down, the same folks give law enforcement credit while leaving the public with a false impression that crime is worse than it really is. This brand of coverage helps explain why the public thinks crime is rising, even when it's falling.
The tactic seemed to nationalize and metastasize after Heather MacDonald began touting her theory of a "Ferguson Effect," which attributed a crime spike predicted based on six months of data to police officers intentionally refusing to do their jobs, effectively as a sort of perverse form of civil disobedience over alleged disrespect they'd received from the Black Lives Matter protests, etc.. Grits would have thought it insulting to claim police officers would intentionally tolerate crime because some protester dinged their ego - it seems like an admission of low integrity to acknowledge that mentality - but somehow MacDonald spun that as a pro-cop view and continues to defend it.
Here's the truth: Nobody can prove she's wrong, but she can't prove she's right. Even if crime does increase when the 2015 numbers come out, what does that mean, and to what should we attribute it? Crime has declined for a couple of decades, but at some point it will reach a floor. When it does, it will rise again. At that point, what sound like scary large increases may not mean much from a statistical perspective. For example, here are Texas' murder rates since 1990 according to annual Uniform Crime Reports the state sends to the feds:
Texas' murder rate declined by 71 percent from 1991 to 2014. (If you want to go back further, our murder rate since 1960 maxed out at 16.9 per 100,000 residents in 1980.)
If Texas' 2015 murder rate jumps to 5.4 per 100,000 people, that would be a 22.7 percent year-over-year increase. Newspaper headlines could and would breathlessly tout a "double digit" increase in statewide murder rates. But the rate would still be 65 percent lower than in 1991.
So when the Dallas Morning News reports an "alarming violent crime increase" because of a few months of data, that's more a tried and true tactic for selling newspapers than an honest assessment of risks to the public. Even sillier is the local union's comment that, “Violent crime is up and murder is up and it’s all because of bad police management.” If violent crime is up, the reasons are far more complex than that. And unless it goes up and stays up, it's not necessarily something about which government or the press should overreact.